Cursor MCP Setup Guide
Cursor MCP Setup Guide
Quick Setup (5 minutes)
Step 1: Verify Everything Works
cd /Users/nickjackolaisen/Documents/GitHub/BI/mcp-servers/government-data
python3 test_client.py
You should see: ✅ Test complete! Client is working.
Step 2: Configure Cursor
- Open Cursor Settings
- Press
Cmd + ,(Mac) orCtrl + ,(Windows/Linux) - Or go to: Cursor → Settings → Features → Model Context Protocol
- Press
-
Add MCP Server Configuration
Find the MCP servers section and add:
{ "mcpServers": { "government-data": { "command": "python3", "args": [ "/Users/nickjackolaisen/Documents/GitHub/BI/mcp-servers/government-data/server.py" ], "env": { "CENSUS_API_KEY": "541100b6baa56ca3bafeffa63e59e80ba1f10d34" } } } }OR if Cursor uses a config file, edit:
~/.cursor/mcp.json(if it exists)- Or add to Cursor’s settings JSON
- Restart Cursor
- Close and reopen Cursor completely
Step 3: Test It!
In Cursor, try asking:
"Fetch business startup data for 2023 from the Census BDS API"
Or:
"Get population data for California from the ACS API for 2023"
The AI should now be able to call the MCP tools automatically!
Troubleshooting
“MCP server not found”
- Check the path is correct:
/Users/nickjackolaisen/Documents/GitHub/BI/mcp-servers/government-data/server.py - Make sure
python3is in your PATH:which python3
“CENSUS_API_KEY not set”
- The API key is in the
.envfile, but Cursor might need it in theenvsection of the config - Add it to the
envsection in the MCP config (see Step 2)
“Module not found: mcp”
- Run:
pip3 install -r requirements.txtin the mcp-servers/government-data directory
Server starts but tools don’t work
- Check Cursor’s MCP logs/console
- Try running the server manually:
python3 server.py(it should wait for input, that’s normal)
What You Can Do Now
Once configured, you can ask the AI to:
- ✅ “Create a stats page about business startups in 2023”
- ✅ “Get business formation data for California”
- ✅ “Fetch job creation statistics for 2020-2023”
- ✅ “Create a comparison of startup rates by state”
The AI will automatically:
- Call the MCP tools to fetch real data
- Use that data to create content
- Follow your existing content patterns
No more manual script running! 🎉